a SpringerOpen Journal

Dida et al. SpringerPlus (2015) 4:136 DOI 10.1186/s40064-015-0905-y

RESEARCH Open Access

Presence and distribution of mosquito larvaepredators and factors influencing their abundancealong the Mara River, Kenya and TanzaniaGabriel O Dida1,2*, Frank B Gelder3, Douglas N Anyona4, Paul O Abuom4, Jackson O Onyuka1, Ally-Said Matano5,Samson O Adoka1, Canisius K Kanangire5, Philip O Owuor6, Collins Ouma1 and Ayub VO Ofulla1

Abstract

Among all the malaria controlling measures, biological control of mosquito larvae may be the cheapest and easiestto implement. This study investigated baseline predation of immature mosquitoes by macroinvertebrate predatorsalong the Mara River, determined the diversity of predators and mosquito larvae habitats and the range of theiradaptive capacity to water physico-chemical parameters. Between July and August 2011, sampling sites (n=39)along the Mara River were selected and investigated for the presence of macroinvertebrate predators and mosquitolarvae. The selected sampling sites were geocoded and each dipped 20 times using standard mosquito larvaedipper to sample mosquito larvae, while a D-frame dip net was used to capture the macroinvertebrate predators. Waterphysico-chemical parameters (dissolved oxygen, temperature, pH, conductivity, salinity and turbidity) were taken in situat access points, while hardness and alkalinity were measured titrimetically. The influence of macroinvertebrate predatoroccurrence was correlated with mosquito larvae and water quality parameters using Generalized Linear Model (GLM).Predators (n=297) belonging to 3 orders of Hemiptera (54.2%), Odonata (22.9%) and Coleoptera (22.9%), and mosquitolarvae (n=4001) belonging to 10 species, which included An.gambiae s.l (44.9%), Culex spp. (34.8%) and An. coustanicomplex (13.8%), An. maculipalpis (3.6%), An. phaorensis (1.2%), An. funestus group (0.5%), An. azaniae (0.4%), An. hamoni(0.3%), An. christyi (0.3%), An. ardensis (0.08%), An. faini (0.07%), An. sergentii (0.05%) and 0.05% of Aedes mosquito larvaewhich were not identified to species level, due to lack of an appropriate key, were captured from different habitatsalong the Mara river. It was established that invasion of habitats by the macroinvertebrate predators were partiallydriven by the presence of mosquito larvae (p < 0.001), and the prevailing water physico-chemical parameters (DO,temperature, and turbidity, p <0.001). Understanding abiotic and biotic factors which favour mosquitoes andmacroinveterbrate co-occurrence may contribute to the control of malaria.

Keywords: Coleoptera; Fish; Hemiptera; Mara river; Mosquito larvae; Odonata; Predators

IntroductionLike in many other parts of the sub-Saharan Africa, mal-aria is increasingly becoming a major health problemamong communities living within river basins includingthe Mara River basin, which stretches between the MaasaiMara game reserve in Kenya and Serengeti National Parkin Tanzania (Bussman et al. 2006). Malaria is now the

* Correspondence: gdidah@gmail.com1School of Public Health and Community Development, Maseno University,Kisumu, Kenya2Department of Vector Ecology and Environment, Institute of TropicalMedicine (NEKKEN), Nagasaki University, Nagasaki, JapanFull list of author information is available at the end of the article

© 2015 Dida et al.; licensee Springer. This is anAttribution License (http://creativecommons.orin any medium, provided the original work is p

leading cause of morbidity and mortality among childrenin many districts within the Lake Victoria basin, such asTrans Mara District, which falls within the Mara Riverbasin of Kenya, while in Tanzania, the disease is commonin almost all regions; the Maasai Mara game reserve beingclassified as low to moderate malaria epidemic area in EastAfrica (Schlagenhauf-lawlor and Scott 2001). According tothe Serengeti Mara Camp Fact sheet, 2013, the famousSerengeti National Park in Tanzania also falls within amalaria endemic zone.One of the most common strategies to eradicate mal-

aria has always focused on mosquito control by use ofvarious chemicals including insecticides. However, their

Open Access article distributed under the terms of the Creative Commonsg/licenses/by/4.0), which permits unrestricted use, distribution, and reproductionroperly credited.

Dida et al. SpringerPlus (2015) 4:136 Page 2 of 14

use have had a negative impact on non-target organismsand the environment. Studies have also shown that someof the chemicals used kill natural mosquito predatorsmore effectively than the target mosquitoes and overtime, predators such as fish and insects die out whilemosquitoes develop resistance, multiplying in ever largernumbers in a losing battle often referred to as “the pesti-cide treadmill” (Wilson and Tisdell 2001). Moreover, theapplication of insecticide strategies has also failed due tothe development of insecticide resistance and lack ofknowledge about the behavior of the vectors (Shiff 2002,Kawada et al. 2011, 2014). The non-selective nature anduse of pesticides therefore leaves biological control of mos-quito larvae as among the best and most environmentally-friendly option for the control of mosquitoes.The role of predatory aquatic insects in the natural

regulation of mosquito larvae has been reported bymany researchers (Knight et al. 2003; Tuno et al. 2005;Mogi 2007; Quiroz-Martinez and Rodriguez-Castro 2007;Shaalan et al. 2007). However, predators vary markedly inthe different habitats that immature and adult mosquitofrequent. Representatives from at least six insect orders,thirteen arachnid families, as well as crustaceans, amphib-ians, fish, birds and mammals have been reported as beingpotential mosquito larvae predators (Mogi 2007, Medlockand Snow, 2008). Some studies in Kenya reported mos-quito larva predation in rice irrigation schemes andwetlands around Lake Victoria (Mwangangi et al. 2007;Minakawa et al. 2007; Ohba et al. 2010). However, most ofthese studies limit their research to specific types ofaquatic insects, principally the Family Notonectidae (Koi-visto et al. 1997; Murdoch et al. 1984). In western Kenya,especially around Lake Victoria, members of the Anoph-eles gambiae s.l. and An. funestus dominate (Minakawaet al. 2008; Minakawa et al. 2012; Kweka et al. 2013).Past experimental studies confirmed that predation on

immature mosquitoes by macroinvertebrates can be amajor driving force in controlling the population size ofmosquitoes, especially the malaria vectors. For instance,Chandler and Highton (1977), reported how predationon An. gambiae larvae resulted in the reduction of thepopulation of the vectors considerably by between 13.4%and 84.5%, respectively. An overall larval mortality ofbetween 92.6% and 97.1% was also reported by Service(1971), (1973) and (1977). Different fish species havealso proved to be effective in mosquito larvae control.However, little research has focused on the assessmentof the available predators’ local ecology to establish theirimpact on mosquito population. Several factors are,however, known to affect the predator-prey relationship.They include preference or selectivity of the prey by thepredator, species diversity in mosquito breeding sites,stability of the aquatic system, larval density, position ofthe predator in the water column, and predator to prey

ratio for the selected micro-environment. Predator-prey co-evolution, predator-prey synchronization andrefuge are also important contributing factors (McPeekand Miller 1996). Environmental factors includingtemperature (Anderson et al. 2001), dissolved oxygen,conductivity (Spieles and Mitsch 2000), and pH(Adebote et al. 2008) may affect predator and preynumbers. From these studies it has been proposed thatfluctuating abiotic conditions and interactions among spe-cies affect predators and prey differentially (Andersonet al. 2001). The physico-chemical differences betweenmosquito and predator breeding habitats are poorlyunderstood and little effort has been made to understandhow these factors affects the vector and prey populationin a shared habitat.The effective control of malaria through vector man-

agement requires information on the distribution andabundance of vectors, as well as factors that favour theiradaptation in the targeted areas. Mosquito larval controlis one potentially important target point in malaria vec-tor control (Kumar et al. 2008; Kweka et al. 2011). Un-derstanding each species biological limits to abioticfactors as well as their habitat structure across environ-mental gradients may provide useful insight into how as-semblages of mosquitoes and mosquito predators arestructured. This information then become useful forproper application of biological control of mosquito lar-vae. Save for a few studies that have been carried out inthe laboratory with species of Anopheles and Culex(Tuno et al. 2006, 2007), little is known on the predationof mosquitoes from rivers and streams in Kenya.This study was therefore designed to determine the pres-

ence and distribution of mosquito larvae predators alongthe Mara River basin, Kenya and Tanzania, for future plan-ning of intervention strategies against malaria and othermosquito-borne diseases through biological control. An at-tempt was also made to further understand the role ofwater physico-chemical parameters on habitat stability.The mean range requirement of physico-chemical parame-ters in habitats shared by the mosquito larvae and preda-tors under field/natural conditions was analyzed from thedata collected within 2 months from 39 sampling sitesalong the Mara River. The identification of indigenouspredator populations and their tolerance to environmentalvariables may help curtail the insurgent of the vector mos-quitoes if a predator propagation program can be initiated.

Materials and methodsStudy areaThis cross-sectional study was conducted at the trans-boundary Mara River basin which lies between longi-tudes 33047’E and 35047’E and latitudes 0038’S and1052’S, traversing Kenya and Tanzania, in East Africa(Mutie et al. 2006). The basin has a tropical rainforest

Dida et al. SpringerPlus (2015) 4:136 Page 3 of 14

climate with two distinct seasons; the rainy season oc-curring between March and May, and the dry season be-tween June and October (Mati et al. 2005). Rainfall inthe basin varies with height, ranging from 1,000 to 1,750mm in the Mau Escarpment, to 900 and 1,000mm in themiddle rangelands and 700 to 850mm in the lower Loitahills and around Musoma in Tanzania, where the riverdischarges into Lake Victoria. The dominant land useswithin the basin are agriculture, pastoralism and wildlifesanctuaries (Mati et al. 2008).

Sampling designSample collection was carried out from the beginning ofJuly 2011 to end of August 2011 (dry season) from 39sampling sites along the Mara River. This study periodwas selected as it presents some of the most extreme en-vironmental variables throughout the year. The specificsampling points were chosen within a 100m intervalfrom one end of the river bank to the other. The siteswere coded based on their location and point of sam-pling. These points were at times strategically chosen be-fore and after a bridge or a through road (for ease ofaccess of both sides of the river), and thus the samplingsites on either side of the bridge or road were labeledsystematically with the first letters denoting their loca-tion as either being upstream or downstream part of theriver, while taking the bridge as the reference point (i.e.URS1-10 and DRS1-10). For instance, URS 1-10 denotedthat the sampling sites 1 to 10 were located on the up-stream side of the river or its tributaries before a bridge,while DRS 1-10 were located on the downstream sideafter the bridge. Other habitats adjacent were givenLatin numbers with their nature and/or name of thehabitat described in detail. This kind of labeling wasdone to avoid any confusion and for easier analysis ofthe specimen. The geographical locations of all the sam-pling sites were taken at access points and recorded asGPS co-ordinates. At every site, all probable habitattypes found were recorded, classified and inspected forthe presence or absence of mosquito larvae and theirpredators. Each habitat was dipped 20 times using astandard mosquito dipper (350 ml; BioQuip Products,Rancho Dominguez, USA). A D-frame dip net of 0.3mwidth attached to a long pole and with a cone-shapedbag for capturing the mosquito larvae predators wasused. The sampling was done from upstream to down-stream end of the river. A total of 3 collections weremade at each sampling point with the collection consist-ing of a forceful thrust of the sampler into the sedimentfor a linear distance of 0.5m. The captured mosquito lar-vae and predators were immediately preserved in 90%ethanol for further identification. Water physico-chemicalparameters (DO, pH, temperature, conductivity, turbidityand salinity) were measured in situ by use of a hand-held

multi-parameter-YSI meter (YSI Model 650-01m Environ-mental Monitoring Systems, Yellow Springs, OH), whilehardness and alkalinity were measured titrimetically.An electro-fisher was used to sample fish only from

the Mara River tributaries of Amala and Nyangorestributaries and the main Mara River. Five sites werechosen randomly and their coordinates taken to repre-sent sampling points in both Kenya and Tanzania. Fishsamples were obtained as per methods described byMatano (2013). Briefly, an electro-fisher that uses apulsed current was used for fishing. This tool was con-nected to an external generator that powered it to pro-duce electric current. The fishing duration at eachstation lasted approximately 30 minutes and covered adistance of 50m and a width of about 3m. Fish sampledwere identified to species level using morphometric andmeristic characteristics following descriptions given byWitte and Van Oijen (1990) and Greenwood (1981).Water physico-chemical parameters were also taken toestablish their influence on fish abundance. Data fromthese sites were however not included in the final modelfor the macroinvertebrates and mosquito analysis as theyfailed to represent all sites and the procedure for sam-pling was also non-selective.

Sampling pointsSampling points (n = 39) were surveyed along the MaraRiver and its tributaries. The sampling points comprisedof macro-habitats including: river (n = 10), drying stream(n = 10), swamps (n = 8), open puddles (n = 5), rock pools(n = 6), dam sites (n = 4), hoof prints (n = 12), vegetatedpools (n = 17) and drainages (n = 25). The remaining sites(n = 29) were mainly open sun-lit pools such as brick-making sites and drainages associated with agricultural ac-tivities in the ephemeral habitats adjacent to the river.

Laboratory analysis and identification of mosquito larvaeand the predatorsAll the collected mosquito larvae were identified micro-scopically using standard taxonomic methods (Gilliesand Coetzee 1987). During the sampling process, therelatively large macroinvertebrate predators such asAnisops wakefieldi (back-swimmers), Rhantus larvae(diving beetles) and Onychohydrus hookeri (water bee-tles) were visually observed, classified and counted.Those that could not be identified to the species level inthe field, including fish were preserved for further iden-tification using appropriate keys as described by Jenkins(1964), Merritt and Cummins (1996), Nilson (1996,1997), Verschuren (1997), Anderson et al. (2001), and byuse of lists of species commonly present in Kenya as de-scribed by Johanson (1992) and Mathooko (1998). Thenumber of mosquito larvae predators was recorded foreach sampled habitat.

Dida et al. SpringerPlus (2015) 4:136 Page 4 of 14

Statistical analysisThe mean differences in water physico-chemical param-eters per habitat types were compared using One-wayANOVA, while the relationship between predators andmosquito larval abundance and the water physico-chemical parameters was determined using the general-ized linear model with negative binomial in MASSpackage, and log as defunct link function. In the initialsteps of the analysis, all the variables were first exploredfor their distribution and the homogeneity of variancechecked using histograms and dot charts after whichthe most appropriate link function was chosen. Theinitial model was built around the premise that the dis-tribution of the response variable was Poisson, whoseover-dispersion was evaluated, and when the condi-tional variance exceeded the conditional mean, a gen-eralized linear model with negative binomial wasemployed. It is considered as a generalization of Pois-son regression since it has the same mean structure asPoisson regression and has an extra parameter tomodel the over-dispersion. A full model included allrelevant covariates, which was then simplified until thebest model with smallest Akaike Information Criterion(AIC) obtained following the stepwise removal of thecovariates. The final model was built as follows; for-mula=(Predators’ abundance) ~ Total mosquito larvae+ Dissolved oxygen + Temperature + Turbidity + pH,(family=GLM.nb). The mean range of water physico-chemical parameters requirements by both mosquitolarvae and their predators in the same habitats wereevaluated using Canonical Correlation Analysis (CCA),as described by Knapp (1978), Härdle and Simar (2007)and Skourkeas et al. (2010), while the contribution ofeach variable in the shared habitats was determinedusing Ordination Analysis (OA). Reliability coefficientof the physico-chemical parameters on fish abundancewas determined using Pearson’s Rank Order Correl-ation Analysis. All statistical analyses were performedusing R (R Core Team, 2013). An alpha value (p < 0.05)was considered statistically significant.

ResultsSampling sites and predator distribution along the MaraRiver and its tributaries are shown in Figure 1 andFigure 2, respectively. A total of 297 macroinvertebratepredators belonging to 3 orders–Hemiptera (54.2%),Odonata (22.9%) and Coleoptera (22.9%) were collected(Table 1). Seven families were recorded within theOrder Hemiptera, with members of Family Velidae andgenus Rhagovelia dominating. Three families were regis-tered within Odonata, dominated by Family Coenagrioni-dae, while Order Coleoptera had 2 families dominated byDytiscidae (Table 1). In addition, a total of 4001 mosquitolarvae were recorded belonging to 10 species, which

included An.gambiae s.l (44.9%), Culex spp. (34.8%) andAn. coustani complex (13.8%), An. maculipalpis (3.6%),An. phaoroensis (1.2%), An. funestus group (0.5%), An.azaniae (0.4%), An. hamoni (0.3%), An. christyi (0.3%), An.ardensis (0.08%), An. faini (0.07%), An. sergentii (0.05%)and 0.05% of Aedes mosquitoes which were not identifiedto species level due to lack of an appropriate key. Themosquito larvae were mainly collected in drying stream,swamps vegetated puddles and open water pools. The ma-jority were collected in drying stream where predatorswere also dominant. The macroinvertebrate predatorsfrom three genera were more abundant where mosquitolarvae were present (Table 2). The distribution of preda-tors and mosquito larvae per habitat type is as presentedin Figure 3. Mosquitoes were captured in the followinghabitats: drying stream (40.1%), swamps (20.0%), vegetatedpools (16.4%), dam (12.4%), open puddles (7.8%), livestockhoof-prints (2.0%) and rock pools (1.3%), among others.Nine species of fish (n=140) representing 4 families

(Cyprinidae, Cichlidae, Claridae, and Poecillidae) werecaptured and identified in the five sampling sites, two onthe main river and three on its tributaries. Cyprinidswere the most abundant, with Barbus altianalis andLabeo victorianus being the most dominant (Figure 4).The most widely distributed species were Barbus altianalisand Gambusia spp. occurring in all sampling sites. Thesewere followed by Labeo victorianus that was present in twosampled sites. Overall, the majority of the fish were caughton the main Mara River (60.9%) as compared to the MaraRiver tributaries; Nyangaores (20.3%) and Amala (18.8%).Except for hardness and salinity, correlations between

six other physico-chemical variables and fish abundancewere evident at all the five sites but the number andstrength of correlations clustered in distinct sites. For in-stance, the strength (significance level) of associationswas greater at upstream sites. Dissolved oxygen andtemperature correlated strongly with fish abundance atsites 1, 2 and 3 whereas sites 4 and 5 although character-ized by relatively swift flow rate, fewer significant associ-ations were recorded (Table 3).Along the Mara River, dissolved oxygen varied consid-

erably among the breeding sites in which the macroinver-terbrate predators and mosquito larvae were caught, withthe highest DO recorded in the river (6.4 ± 0.7 mg/L),followed by rock pools (6.0 ± 0.7 mg/L). The lowest wasrecorded in swamps (2.4 ± 2.7 mg/L). The overall meanDO in puddles was (5.6 ± 0.8 mg/L), while that of dryingstream was (5.3 ± 1.6 mg/L) (Table 4). A significant differ-ence in mean DO was observed among the 9 different habi-tat types (ANOVA, n = 9, F = 4.2417, d.f. = 8, 26, p < 0.01).It was established that both mosquito larvae and macroin-verterbrate predators along the Mara River were prevalentin samples with DO values ranging between 6.0 mg/L, to6.5 mg/L.

Figure 1 Red dots show the sampling sites along the Mara river and tributaries, Kenya and Tanzania (n = 39).

Hemiptera

Coleoptera

Odonata

1

2

4

6

Average predator abundance Predator Order

Figure 2 Dot size and color show predator order and average number at sampling sites along the Mara river and tributaries, Kenyaand Tanzania (n = 39).

Dida et al. SpringerPlus (2015) 4:136 Page 5 of 14

Table 1 Order, family, genus, number and percent (%) forall of mosquito larvae predators captured throughoutthis study at all locations

Order (n) Family(common name)

Genus/species(sub-order)

n (%)

Hemiptera (161) Gerridae Hynesionella(Nepomorpha)

7 (2.4)

Limnogonus(Gerromorpha)

13 (4.4)

Hydrometridae Hydrometra species 15 (5.1)

Veliidae Rhagovelia(Heteroptera)

38 (12.8)

Notonectidae Anisops (Anisoptera) 30 (10.1)

Enithares species 9 (3.0)

Pleidae (Waterbug)

Pleidae species 8 (2.7)

Naucoridae Naucoridae species 7 (2.4)

Nepidae Ranatra species 10 (3.4)

Laccotrephes 24 (8.1)

Odonata (68) Lestidae(Damselfly)

Lestes species 20 (6.7)

Coenagrionidae Enallagma species 21 (7.0)

Libellulidae Palpopleura 14 (4.7)

Orthetrum albistylum 13 (4.4)

Coleoptera (68) Hydrophilidae(Water beetle)

Hydrocharacaraboides

8 (2.7)

Dytiscidae Laccophillus species 49 (16.7)

Copelatus species 4 (1.3)

Cybister species 6 (2.0)

Hydaticus species 1 (0.3)

TOTAL (N) 297(100)

Table 2 Mosquito larvae and predator numbers indifferent habitats within the Mara river basin

Habitat Mosquito(n)

Proportion(%)

Predators(n)

Proportion(%)

Drying stream 1009 25.2 120 40.4

Swamps 830 20.7 92 31.0

Open puddles 524 13.1 4 1.4

Dams 510 12.8 13 4.4

Vegetated pools 455 11.4 45 15.2

Hoof prints 250 6.3 4 1.4

Drainages 234 5.8 13 4.4

Rock pools 188 4.7 3 1.0

River 1 0.0 3 1.0

TOTAL (N) 4001 100 297 100

Dida et al. SpringerPlus (2015) 4:136 Page 6 of 14

Conductivity levels across different habitats showedwide variation, ranging between a mean of 144.5 ± 97.6μS/cm for the rivers and 368.0 ± 125.9 μS/cm for therock pools. Dams and drying stream habitats also re-corded relatively high mean conductivity levels of be-tween 269.8 ± 213.8 μS/cm and 290 ± 186.5 μS/cm,respectively. Measurements from drainages (168.5 ± 13.4μS/cm), open sunlit puddles (168.8 ± 87.3 μS/cm),swamps (174.3 ± 59.2 μS/cm), dams (269.8 ± 213.8 μS/cm) and drying stream (290 ± 186.5 μS/cm) demon-strated marked variation in mean values. The lowestmean values were recorded in river habitats with mea-sured mean ranges of 155.7 ± 88.4 μS/cm and 144.5 ±97.6 μS/cm, respectively (Table 4). ANOVA test revealeda significant difference in electrical conductivity amongthe habitat types (ANOVA, n = 9, F = 7.1433, d.f.=8, 26,p < 0.01). Conductivity requirement range by both mos-quito larva and predators varied markedly between 109.9μS/cm to 396.2 μS/cm. However, ranges between 162.9μS/cm and 166 μS/cm were most preferable based onmosquito larva and predator numbers captured in theshared habitats along the Mara River and its tributaries.Far fewer mosquito larvae and predators were present insamples at the extremes of these measurements.Water pH measurements varied markedly between dif-

ferent habitats, ranging between 6.7 to 8.4. The highestmean value was recorded in open puddle habitats (8.2 ±0.5), while the lowest (7.0 ± 1.3) was recorded inswamps. Drying stream and dam water pH measure-ments were comparable at 8.1 ± 0.6 and 8.1 ± 0.4, re-spectively. In drainages, the mean pH value was 7.3 ±0.5, while river had mean of 7.3 ± 0.4. Rock pools, animalhoof prints and vegetated pools had mean pH values of7.1 ± 0.8, 8.1 ± 0.3 and 8.0 ± 0.2, respectively (Table 4).There was significant differences in mean pH among thehabitat types (ANOVA, n = 9, F = 9.443, d.f. = 8, 26,p < 0.01). Both mosquitoes and their predators werehowever abundant in the pH range of between 6.7 to8.4, respectively.Water temperature changes are influenced by many

variables including time of sampling, source of thewater and condition of the habitat. Along the MaraRiver, the highest mean temperature was recorded inthe rock pools (26.2 ± 3.4°C), and animal hoof-prints(26.2 ± 1.9°C), followed by puddles (25.2 ± 2.3°C). Riversamples had the lowest temperature (19.7 ± 2.3°C).Temperatures in dam (24.4 ± 1.9°C), drainages (24.2 ±0.7°C), swamps (23.2 ± 4.9°C) and drying stream (24.4± 1.9°C) varied slightly during the study period.Overall, the temperature variation within the MaraRiver and its tributaries as tested between July andAugust ranged from 18.0°C to 26.3°C. A significant differ-ence in mean temperature was observed among thedifferent habitat types (ANOVA, n = 9, F = 4.2004, d.f = 8,

Figure 3 Abundance of macoinvertebrate predators by order along the Mara river and its tributaries, Kenya and Tanzania, n = 39.

Dida et al. SpringerPlus (2015) 4:136 Page 7 of 14

26, p = 0.05). Both predator and prey preferred tempera-tures above 18°C. In particular, temperatures above 25°Ccontained the greatest number of both predator and prey.At these temperatures, some pool samplings recordedhigh predator numbers with very few prey, suggesting suc-cessful predation.As presented in Table 4, the highest mean alkalinity

(400 ± 282.8 mg/L) was recorded in the drainages whilethe lowest were recorded in dams and driver (100 ± 62.4mg/L and 100 ± 99.2 mg/L), and vegetated pools (104 ±73.0 mg/L). Similarly, variations between drying stream(126.2 ± 26.5 mg/L), puddles (104 ± 73.0 mg/L), rockpools (153 ± 60.8 mg/L), swamps (244.5 ± 274.6 mg/L)and animal hoof prints (133 ± 50.2 mg/L) were deter-mined. Mean water alkalinity values differed significantlybetween habitat types along the Mara River (ANOVA,n = 9, F = 4.7042, d.f. = 8, 26, p < 0.001). Alkalinity rangerequirement for both mosquito larvae and predators inthe shared habitats varied, with values ranging between6.4 mg/L, and 406.1 mg/L. The most abundant collec-tions of both larvae and their predators were in the

6.3

4.7%

1.2%

0.0 5.0

Barbus altianalis

Labeo victorianus

Gambusia spp.

Oreochromis leucostictus

Clarias liocephalus

Barbus neumayerii

Clarias gariepinus

Barbus nyanzae

Haplochromine spp.

Fish

spe

cies

Figure 4 Fish species sampled in the Mara river and its tributaries, n

range of 131.2 mg/L and 144.4 mg/L. From the abovedata, neither mosquito larvae nor predators had specificalkalinity requirement. Further analysis to determine thepreferable alkalinity range requirement by both mos-quito larvae and predators in the shared habitats indi-cated that only few insects preferred a range between131.2 mg/L, and 144.4 mg/L, majority had a more widerrequirement range. Similarly, this study found no spe-cific preferences for hardness. For salinity, only swampsrecorded slight salinity of 0.4 mg/L, while all the othersites recorded zero (Table 4).A negative binomial GLM results established that the

abundance of the predators in habitats were partially drivenby the presence of mosquito larvae (Z = 6.49, p < 0.001),and the prevailing water physico-chemical parameters (dis-solved oxygen, Z = 3.34, p < 0.001; temperature, Z = 2.75,p < 0.001; and turbidity, Z =-3.65, p < 0.001), based on thebest model with the smallest AIC (Table 5). To evaluate thestrength and pattern of relationship between mosquito lar-vae and macroinverterbrate predators, a canonical correl-ation analysis was done. There was a strong correlation

23.0%

17.4%

15.8%

14.2%

9.0%

8.4

%

10.0 15.0 20.0 25.0

Abundance

= 5.

Table 3 Spearman rank order correlation results forassociation between fish abundance and the physico-chemical parameters at the Mara River and tributaries, n = 5

Variable Site 1 Site 2 Site 3 Site 4 Site 5

Dissolved Oxygen 0.65*** 0.62*** 0.44** − −

pH 0.52** 0.28* − − −

Conductivity 0.24* − 0.27* − −

Turbidity 0.38** − − − −

Temperature 0.66*** 0.74*** − 0.18 0.38**

Hardness − − − − −

Salinity − − − − −

p <0.1, *p <0.05 **p <0.001 ***p <0.005; denotes strengths of correlation atdifferent sites. Dissolved oxygen and temperature correlated strongly with fishabundance at sites 1, 2 and 3.

Dida et al. SpringerPlus (2015) 4:136 Page 8 of 14

between the predators and mosquito larvae (R2=0.62,p < 0.005). Data from some sites showed inverse correlationbetween predators and prey (mosquito larvae), suggestingeffective predation (Figure 5). The biplot (Figure 6), with in-tuitive interpretations of species-biotic interaction of all 9variables confirmed that pH, alkalinity and hardness wereless likely to influence mosquito larvae (An. gambiae com-plex and Culex spp.) and predators abundance. Dissolvedoxygen and temperature were the most important factorsthat positively and directly correlated with both mosquitolarvae and predators abundance based on quadrant reflec-tion in the ordination analysis.

DiscussionMalaria is a preventable and curable disease when en-countered under ideal circumstances. However, underless ideal circumstances, or in regions where malaria isendemic such as the Sub-Saharan Africa, malaria mor-bidity and mortality continues to result in human andeconomic disability (WHO 2013). Resistance to chemicalinsecticides during the late 1950s resulted in an expected

Table 4 Average physico-chemical parameters at different mo

Habitat DO (mg/L) pH Alkalinity(mg/L)

Hardness(mg/L)

Dams 4.7 ± 1.8 8.1 ± 0.4 100 ± 62.4 87.7 ± 56.2

Drying stream 5.3 ± 1.6 8.1 ± 0.6 126.2 ± 26.5 102.4 ± 68.9

Swamps 2.4 ± 2.7 7.0 ± 1.3 244.5 ± 274.6 58.5 ± 46.7

Drainages 4.3 ± 3.8 7.3 ± 0.5 400 ± 282.8 372 ± 393.2

Rock pools 6.0 ± 0.7 7.1 ± 0.8 153 ± 60.8 127 ± 69.3

Open puddles 5.6 ± 0.8 8.2 ± 0.5 104 ± 73.0 188 ± 247.7

River 6.4 ± 0.7 7.3 ± 0.4 100 ± 99.2 178 ± 228.8

Hoofprints 6.2 ± 0.5 8.1 ± 0.3 133 ± 50.2 98.9 ± 46.5

Vegetated pools 5.4 ± 0.6 8.0 ± 0.2 120 ± 72.5 104.1 ± 98.8

*Elevated levels of turbidity and conductivity were recorded in rock pools, probably

turn toward a search for biocontrol agents against themosquito larvae. Some organisms are more chemical tol-erant than others, and aquatic insects are sensitive tochange of their environment. For instance, spraying ofpesticides in agricultural fields along the river channelhas been reported to have negative consequences onaquatic insects by Gereta et al. (2003). Therefore,alternative malaria control strategy of bio-environmentalimprovement techniques gives primary importance toanti-larval operations. Drying streams supported thegreatest numbers of both mosquito larvae and predatorsduring this sampling period and may be responsible forincreasing natural predation in certain temporary habi-tats such as dams, open puddles and vegetated pools.This possibility is supported by the observation that cer-tain ephemeral aquatic habitats had lower number ofmosquitoes and higher predator abundance.Water temperature, turbidity, and dissolved oxygen were

found to be the main variables influencing the abundanceand distribution of mosquito larva and predators in theaquatic habitats under investigation, even as test suggestthat they opt more for clearer water. Within these habitats,ephemeral aquatic habitats had a diverse array of preda-tors, which in some instances, correlated with negligiblemosquito larvae numbers. These habitats results primarilydue to human settlements. Brick making, cultivation, stray-ing wildlife from the adjacent Masai Mara National Parkand keeping of livestock have created animal hoof-prints.Consequently, open puddles, drainages and hoof-printsfound in these areas supported a considerable number ofmosquito and mosquito larvae predators.The order Hemiptera were the most dominant and

widespread representing 7 families. The 7 families wereover-represented by Family Velidae and Genus Rhagovelia.Other predators of mosquito larvae belonged to the OrderOdonata (which recorded 3 families dominated by familyCoenagrionidae) and Order Coleoptera (which recorded 2

squito larvae habitats along Mara the River basin

Turbidity (NTU) Conductivity(μS/cm)

Temperature (0C) Salinity(mg/L)

96.9 ± 142.0 269.8 ± 213.8 24.4 ± 1.9 0.0 ± 0.0

124.3 ± 152.6 290 ± 186.5 22.5 ± 2.1 0.0 ± 0.0

142.2 ± 108.5 174.3 ± 59.2 23.2 ± 4.9 <0.1

144.8 ± 84.3 168.5 ± 13.4 24.2 ± 0.7 0.0 ± 0.0

542.6 ± 2.3* 368.0 ± 125.9* 26.2 ± 3.4 0.0 ± 0.0

95.2 ± 131.9 168.8 ± 87.3 25.2 ± 2.3 0.0 ± 0.0

135.2 ± 142.4 144.5 ± 97.6 19.7 ± 2.3 0.0 ± 0.0

100.2 ± 62.1 140.3 ± 90.4 26.2 ± 1.9 0.0 ± 0.0

150.2 ± 102.4 135.2 ± 142.4 19.7 ± 2.3 0.0 ± 0.0

due accumulation of dissolved particles.

Table 5 Final nb-GLM model for the response variable(mosquito larvae predators) and their predictors(mosquitoes and the physico-chemical parameters) thatremained in the model, denoting factors influencingmosquito predators abundance in habitats along theMara river

Variable Estimate Std. error z value Pr (>|z|)

Intercept -3.45 1.22 -2.83 0.005

Dissolved oxygen (DO) 0.38 0.11 3.34 <0.001

Temperature 0.07 0.03 2.75 0.006

Turbidity -0.01 0.01 -3.63 <0.001

Mosquito larvae 0.41 0.10 6.49 <0.001

Dida et al. SpringerPlus (2015) 4:136 Page 9 of 14

families dominated by Dytiscidae). The Hemipterans areregarded as effective predators of freshwater snails andmosquito larvae (Ohba and Nakasuji 2006). It is also wellknown that notonectids are voracious predators of mos-quito larvae. Gilbert and Burns (1999) concluded thatnotonectid predators have the potential to alter mosquitocommunities via direct or indirect effects. Direct evidenceof notonectid predation on mosquito larvae was laternoted and this further confirmed their predominant rolein mosquito larvae control (Chesson 1984).The relatively low mosquito and predator numbers ob-

served in the ephemeral habitats as compared to dryingstream and swamps might have been due to several rea-sons. In addition to the fact that most of these habitatsare open and might be accessible by the predators, earl-ier studies also reported that adult mosquitoes may havethe ability to detect presence of predators and com-pletely avoid ovipositing in such habitats, preferring in-stead to inhabit areas with swamps and grassy patchesthat can protect the immature stages (Vince et al. 1976;Nelson 1979; Coen et al. 1981; Heck and Thoman 1981;Stav et al. 1999). Previously, mosquitoes of the speciesCuliseta longiareolata were reported to detect chemicalsfrom notonecta predators, and the instinct/cue can existin the habitat for up to a week or more after their dis-appearance from the pool (Blaustein et al. 2004) and forCulex species, a period as low as two days have been re-ported (Blaustein et al. 2005).We noted that the majority of predators were bonded

to where there were lower densities of mosquito asreflected in graphical multi-correlation matrix. This sup-ports the above aforementioned studies. However, highernumber of predators and less prey could also be as a re-sult of direct predation. It was therefore reasonable toexpect fewer mosquitoes in habitats with higher numberof predators. Other factors that have previously been re-ported to play an important role in habitat selection byvarious species of mosquitoes are volatile compoundsproduced by microbial population in the breeding sites(Sumba et al. 2008), chlorophyll content in the breeding

sites (Munga et al. 2013) and other abiotic factors whichcan inhibit adult mosquitoes oviposition, coupled withhabitat preferences (Minakawa et al. 1999, 2012).The temperature recorded in the current study ranged

between 18.0°C and 26.3°C, thus can be described as warmand are likely to support most of the predators especiallythe notonectids. Earlier studies showed that thermal con-ditions are especially important in predator–prey survivalamong aquatic organisms (Bailey 1989, Thomson 1978),especially those that are involved in size-dependent preda-tion (Formanowicz 1986, Travis et al. 1985). However,while much research quantified in the physiological effectsof temperature on specific organisms, few studies havebeen conducted to evaluate the effect of temperature onspecies interactions and their adaptive capacity to thoseranges in field conditions.Mosquito larvae and predators share the same habi-

tats and thus establishing the role that pH plays in theregulation of colonization is critical. Both mosquito lar-vae and predators were not affected by pH in the finalGLM model. This suggests that under the prevailingenvironmental conditions, both predators and mosqui-toes could tolerate a wide range of pH. Further analysisto determine preferable pH range requirement by bothmosquito larvae and predators established that valuesbetween 6.7 and 8.4 were tolerable, while values be-tween 8.1 and 8.4 were most preferred, as evidenced bythe highest number of both mosquito larvae and preda-tors. The pH was largely basic in all habitat types. Theadaptive range of pH by the insects was wide andwithin that range. Alkalinity levels were equally highranging between 100 and 420 mg/L. This pH range hasbeen reported as optimal for most aquatic biota includ-ing some mosquito predators. Most findings agree withthe positive association of mosquito larvae and otheraquatic insects under a wide range of pH values. For in-stance, Adebote et al. 2008 found mosquitoes of speciesof An. ardensis, An. distinctus, and An. wilsoni to be as-sociated with pools of acidic nature (pH 5.86-6.55);however, Cx. ingrami occurred in partly acidic andpartly alkaline pools (pH 5.86-9.85). Similarly, a studyby Dejenie et al. (2011) on malaria vector control inEthiopia established that almost all their study habitatswere alkaline (pH >7) and both anopheline and culicinelarvae were positively associated with this high (>7.0)pH. Our study thus is in agreement with the study ofDejenie et al. (2011), but does not support the findingsof Adebote et al. (2008), which reported the preferenceof anopheline species to low pH values.Along the Mara River, the mean turbidity was highest in

rock pools, while the lowest level was recorded in swampsand drainages. The findings showed that turbidity levelsacross all sampled sites were exceedingly high. This sce-nario could be as a result of increased particulate matter

Pre

dato

rs a

bund

ance

Number of mosquito larvae

Figure 5 Correlation matrix showing correlation between predators (blue stars) and the mosquitoes (absolute number) in sharedhabitat along the Mara river (n = 39).

Dida et al. SpringerPlus (2015) 4:136 Page 10 of 14

such as clay, silt, organic matter, plankton and other micro-scopic organisms, which have been reported to interferewith the passage of light through water (American PublicHealth Association APHA 1998). The increased particulatematter could have been contributed by anthropogenic ac-tivities such as deforestation, river bank cultivation, soilerosion (due to overgrazing among others), all occurring inthe watershed. In addition, urbanization facilitates trans-portation of waste into the river channel through increasedrun-offs, while livestock trampling effect at watering pointsand along the river banks also contributes significantly tohigh turbidity levels of surface waters. All these activitiescan create suitable habitats for mosquitoes as was previ-ously reported by Matthys et al. (2006).A habitable aquatic ecosystem requires a good supply of

dissolved oxygen in the water system (Davis 1975). Along

Figure 6 Biplot of the overall effect of various environmental parame

the Mara River basin, the mean dissolved oxygen was high-est in the river followed by rock pools, while the lowest wasrecorded in swamps. A significant difference in mean dis-solved oxygen was observed among the different habitattypes. Faster flowing sections of rivers and drying streamand sections that flow through riffles or small waterfallshave better oxygenated waters than slow flowing sections ofrivers or rivers that have been modified as straight channels.Dissolved oxygen concentration in water is dependent onphysical, chemical, biological and microbiological processes.Low dissolved oxygen concentrations (<3 mg/L) in freshwater ecosystems are indicative of high pollution levels(Okbah and Tayel 1999). However, in the current study,some aquatic habitats recorded dissolved oxygen levels in-sufficient to support aquatic life. Analysis to determinepreferable level of dissolved oxygen range required by both

ters recorded along the Mara river (n = 39).

Dida et al. SpringerPlus (2015) 4:136 Page 11 of 14

mosquito larvae and predators in the shared habitat indi-cated that values ranging between 6.0 mg/L, and 6.5 mg/Lwere most preferred. However, some mosquito larvae werefound in water samples with dissolved oxygen concentra-tion as low as 2.3 mg/L. The most common cause of lowoxygen levels is the off-load of organic material into thewater system (such as agricultural run-offs). Nevertheless,more mosquito larvae were collected in slow-flowing dryingstream and swamps where the mean oxygen was relativelylow. The majority were mainly Culex spp., however Anoph-eles species were also higher as compared to the ephemeralhabitats. previous studies reported Culex spp. to occur inhabitats with wide range of dissolved oxygen (DO) levels(Opoku et al. 2007).This may also suggest that majority ofpredators were unable to survive in polluted water or thewater volume was sufficient to maintain high number ofmosquitoes beyond predator’s capacity. Also, streams andrivers have burrows at the banks which can help in refugeof the mosquitoes. In fact, majority of the mosquitoes werecaptured at the edges of the streams, most of which arevegetated.The majority of Anopheles and Culex spp. larvae were

found inhabiting pools adjacent to the Mara River cre-ated by receding river waters, some of which had rela-tively high dissolved oxygen levels. These findings areconsistent with those of Dejenie et al. (2011) who alsoreported that both Anopheline and Culicine larvae werepositively associated with dissolved oxygen. Studies byMuturi et al. (2008) also indicated similar association ofAnopheles spp. larvae and other mosquito larvae withdissolved oxygen. Likewise, Oyewole et al. (2009) em-phasized that optimum dissolved oxygen is superlativeto the survival of the Anopheles larvae.Water hardness is usually a result of the presence of

multivalent metal from minerals dissolved in water. Inthe aquatic environment, ions result from abundance ofcalcium and magnesium in water. The highest meanhardness was recorded in the drainages, while the lowestwere recorded in dams and swamps. A correlationmatrix established that there was a positive correlationbetween mosquito larvae and predators in the presenceof hardness. However, a negative correlation was ob-served between hardness and predators in the sharedhabitats suggesting that most predators require lowerwater hardness levels to survive. Analysis to determinethe preferable level of hardness range requirement byboth mosquito larvae and predators in the shared habitatsindicated that values as wide as 58.5 mg/L to 397.1 mg/L,were favorable. The wide range of water hardness ob-served could be due to differences in buffering capacity ofthe waters across habitat types, as hardness values are notconsistent across the basin. Elevated values in some areascould be as a result of sewer supply from the nearby townsor spills of fertilizer from the nearby farms. Other

established sources could be the local geology (Lawrence2007). However, few insects showed preference for specifichardness values. It was also of interest to note that alongthe Mara River, most aquatic habitats had meagre detect-able level of salinity. Only swamps recorded salinity levelof 0.4 mg/L. However, the influence of salinity along theMara River could not be statistically evaluated as a resultof insufficient sample numbers.In the current study, rock pools, dams and drying

stream recorded the highest mean conductivity, whileswamps and drainages had the lowest conductivityvalues. For both mosquito larvae and predators, a perfectlinear requirement with conductivity in the same habitatwas demonstrated within the ranges of between 162.9μS/cm to166μS/cm by both mosquito larvae and preda-tor residing in the same habitats. The high levels weredue to elevated dissolved solids and contaminants espe-cially electrolytes. Potential sources of these contaminantsare destruction of the forest cover (which in the process,increase the litters) and human activities experiencedalong the river channel (that creates drainages and poolssuitable for mosquito breeding). Mati et al. (2008) andJordao et al. (2007) reported increased destruction of theupper catchment of the Mau forest and elevated level ofpollution, attributable to high levels of waste water dis-charged into the river from different origins.Previously, dissolved oxygen, temperature and con-

ductivity were reported to positively correlate with mac-roinvertebrate community structure as a whole (Spielesand Mitsch 2000). In the current study, no direct rela-tionship was detected between conductivity and preda-tor abundance in the GLM model. However, there was alimited range of conductivity requirement levels prefer-able to both mosquito and predator population. Theconductivity of a river or stream should remain within aspecified range to allow for a successful biologically func-tional system. Changes in conductivity are often used aswater pollution indicator. Urban run-offs and industrialpollution are often characterized by high conductivity.Ordination analysis, factoring in all the variables showed

that dissolved oxygen and temperature had direct influ-ence on mosquito larvae and predator abundance, whileother biotic factors indicated meagre, opposite or insignifi-cant role, supporting the earlier notion that within aquatichabitat, both macroinvertebrates and mosquitoes can besensitive to factors affecting water quality. Gauch (1982)concur that ordination primarily endeavors to representsample and species relationships as faithfully as possible inorder to choose precisely which tool is necessary for im-mediate use. Predator abundance was strongly positivelycorrelated with the increasing number of mosquitoes, sug-gesting that carefully selected predators may play a noblerole in controlling mosquitoes as compared to the waterphysico-chemical parameters. Specifically, the abundance

Dida et al. SpringerPlus (2015) 4:136 Page 12 of 14

of predators, Culex and Anopheles spp. were perfectly cor-related in ordination analysis. pH, turbidity, conductivity,alkalinity, hardness and salinity, whereas some were corre-lated with each other, the higher they were, the less likelythat the mosquito and predators would inhabit, exceptwith the availability appropriate water prysico-chemical re-quirement range. Previous studies reported that thermalpollution, pesticides and organic compounds affects thewater physico-chemical parameters, thus interfering withaquatic invertebrate diversity and composition (Hilsenhoff1988). This may also partially explain the abundance ofHemiptera, as compared to the other two aquatic insectorders; Odonata and Coleoptera. Most of the insects inthe Order Hemiptera have been acclaimed as pollution-tolerant (Joshi 2012), and their population was found tobe higher than any other order along the Mara River.Other known sensitive taxa such as Plecoptera were com-pletely absent from all the sites, suggesting that the watersmight have been polluted.Downstream sites of the Mara River had greater flow

rates, more stable physico-chemical parameters, and poten-tially greater influence from tributaries in the biotic interac-tions. Of interest is the fact that more fish species werepresent in middle and lower Mara River sites as comparedto upstream sites, where most physic-chemicals had ex-treme ranges, and thus the biotic interactions were poten-tially more important in regulating fish abundanceupstream. We speculate that physico-chemical variables areimportant influencers on fish diversity and abundance inthe Mara River. Therefore, future studies determining fishcommunity structure and the role of the biotic factorsshould further elucidate their importance with proper de-sign and robust analysis.

ConclusionsMosquito resistance to chemical insecticides is a growingproblem, and increasing attention is being paid to alterna-tive control methods. The findings reported herein providenew information on the presence of macroinvertebrate andmosquito larvae within the Mara River and its tributaries.Some of these predatory species have been evaluated asbio-control agents in the worldwide campaign to controlmalaria vectors. This study also defines the most preferablephysico-chemical parameter range dependency by the pred-ators and mosquito larvae. Understanding abiotic and bioticfactors which favour mosquitoes and macroinveterbrate co-occurrence, may contribute to the control of malaria.

Limitations of the studyThis study was designed and conducted during the dryperiod along the Mara River tributaries and the mainMara River in Kenya and Tanzania. We presumed thatsamples collected during this dry period would representextremes within this ecosystem. From our data extremes,

multiple variables were measured. However, we recognizethe limitations of a crossectional study in which all aspectsof habitat parameters, especially changes in the physico-chemical parameters over time are not represented. Also,a more extensive fish sampling should have been con-ducted and in a more scientific manner that would ensureconcrete conclusion are made, including generalization ofthe results to larger groups.

AbbreviationGLM: Generalized linear model.

Competing interestsThe authors declare that they have no competing interests.

Authors’ contributionsGD, AO, FG, PO, PA, CK and CO conceived and designed the study. AO, PO, PA,SA, DA, CO and AM organized the field work. GD, SA, JO and DA collected thedata. GD and DA performed the statistical analysis. GD wrote the paper. GD, DA,AO and FG revised the manuscript. All authors read and approved the finalmanuscript.

AcknowledgementsWe thank Mr Gordon Opiyo Okute for his help with mosquito identificationand the field assistants for their help in sample collection. We also thankAdams Schilder for his input and the Kenya Marine and Fisheries ResearchInstitute for allowing the use of their equipment during field work and theirlaboratory for sample analysis. This study was financially supported by theEast Africa Community-Lake Victoria Basin Commission Secretariat.

Author details1School of Public Health and Community Development, Maseno University,Kisumu, Kenya. 2Department of Vector Ecology and Environment, Institute ofTropical Medicine (NEKKEN), Nagasaki University, Nagasaki, Japan. 3ProbeInternational, Inc., USA and Auckland, Ohio, New Zealand. 4School ofEnvironment and Earth Science, Maseno University, Kisumu, Kenya. 5EAC-LakeVictoria Basin Commission Secretariat, Kisumu, Kenya. 6Department ofChemistry, Maseno University, Kisumu, Kenya.

Received: 14 October 2014 Accepted: 24 February 2015

ReferencesAdebote DA, Oniye SJ, Muhammed YA (2008) Studies on mosquitoes breeding in rock

pools on inselbergs around Zaria, northern Nigeria. J Vector Borne Dis 45:21–28American Public Health Association (APHA) (1998) Standard Methods for the

Examination of Water and Wastewater (20th Edition). American Public HealthAssociation, American Water Works Association, and the Water EnvironmentFederation, Washington, D.C

Anderson MT, Kiesecker JM, Chivers DP, Blaustein AR (2001) The direct andindirect effects of temperature on a predator–prey relationship. Can J Zool79:1834–1841, doi: 10.1139/cjz-79-10-1834

Bailey PCE (1989) The effect of water temperature on the functional response ofthe water stick Ranatra dispar (Heteroptera: Nepidae). Aust J Ecol 14:381–386

Blaustein L, Kiflawi M, Eitam A, Mangel M, Cohen JE (2004) Oviposition habitatselection in response to risk of predation in temporary pools: mode of detectionand consistency across experimental venue. Oecologia 138(2):300–305

Blaustein L, Blaustein J, Chase J (2005) Chemical detection of predator Notonectairrorata by ovipositiong Culex mosquitoes. J Vector Ecol 30:299–301

Bussman RW, Gilbreath CG, Lutura M, Lutuluo R, Kunguru K, Wood N, MathengeS (2006) Plant use of the Maasai of Sekenani Valley, Maasai Mara, Kenya. J ofEthnobiol and Ethnomed 2:22

Chandler JA, Highton RB (1977) Breeding of Anopheles gambiae Giles (Diptera:Culicidae) in rice fields in the Kisumu area of Kenya. J Med Entomol13:211–215, PubMed: 978695

Chesson J (1984) Effects of notonectids (Hemiptera: Notonectidae) onmosquitoes (Diptera: Culicidae): Biorational Control of Mosquitoes 115predation or selective oviposition. Environ Entomol 13:531–538

Dida et al. SpringerPlus (2015) 4:136 Page 13 of 14

Coen LD, Heck KL Jr, Abele LG (1981) Experiments on competition and predationamong shrimps of seagrass meadows. Ecology 62(6):1484–1493

R Core Team (2013) R: A language and environment for statistical computing. RFoundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0,URL http://www.R-project.org/

Davis JC (1975) Minimal dissolved oxygen requirements of aquatic life withemphasis on Canadian species: a review. J Fish Res Bd Canada 32:2295–2332

Dejenie T, Yohannes M, Assmelash Y (2011) Characterization of mosquitobreeding sites in and in the vicinity of Tigray microdams. Ethiop J Health Sci21(1):57–66

Formanowicz DR (1986) Anuran tadpole/aquatic insect predator–preyinteractions: tadpole size and prey capture success. Herpetol 42:367–373

Gauch HG (1982) Multivariate Analysis and Community Structure. CambridgeUniversity Press, Cambridge

Gereta EJ, Chiombola EA, Wolanski E (2003) Assessment of the EnvironmentalSocial and Economic impacts on the Serengeti ecosystem of thedevelopment in the Mara River Catchments

Gilbert JJ, Burns CW (1999) Some observation on the diet of the backswimmer,Anisops wakefieldi (Hemiptera: Notonectidae). Hydrobiol 412:111–118

Gillies MT, Coetzee M (1987) A supplement to the Anophelinae of Africa south ofthe Sahara (Afrotropical region). S Afr Inst Med Res 55:1–139

Greenwood PH (1981) The Haplochromine Fishes of the East African Lakes. BritishMuseum (Natural History) and Kraus International Publications, London andMunich

Härdle W, Simar L (2007) Canonical Correlation Analysis. Applied MultivariateStatistical Analysis., pp 321–330, doi:10.1007/978-3-540-72244-1_14.ISBN 978-3-540-72243-4

Heck KL Jr, Thoman TA (1981) Experiments on predator-prey interactions invegetated aquatic habitats. J exp. mar. Biol Ecol 53:125–134

Hilsenhoff WL (1988) Rapid field assessment of organic pollution with a family-level biotic index. J North Am Benthol Soc 7(1):65–68

Jenkins DW (1964) Pathogens, parasites and predators of medical importancearthropods: annoted list and bibliography. Bull Wld Hlth Org 30(Suppl):150

Johanson KA (1992) A catalogue of the caddisflies of East Africa (Insecta,Trichoptera). Steenstrupia 18:113–141

Jordao CP, Ribeiro PRS, Matos AT, Bastos RK, Fernandes RBA, Fontes RLF (2007)Environmental assessment of water-courses of Turvo Limpo River basin atthe Minas Gerais state, Brazil. Environ Monit Assess 127:315–326

Joshi PP (2012) Aquatic hemipteran diversity as indicators of more environmentalextremes: relation to tolerant of some physico-chemical characteristics ofwater. Biosci Disc 3(1):120–124

Kawada H, Dida GO, Ohashi K, Komagata O, Kasai S, Tomita T, Sonye G, MaekawaY, Mwatele C, Njenga SM, Mwandawiro C, Minakawa N, Takagi M (2011)Multimodal pyrethroid resistance in malaria vectors, Anopheles gambiae s.s.,Anopheles arabiensis, and Anopheles funestus s.s. in Western Kenya. PLoS One6:e22574

Kawada H, Ohashi K, Dida GO, Sonye G, Njenga SM, Mwandawiro C, Minakawa N(2014) Insecticidal and repellent activities of pyrethroids to the three majorpyrethroid-resistant malaria vectors in western Kenya. Parasites & Vectors7:208

Knapp TR (1978) Canonical correlation analysis: A general parametric significance-testing system. Psychol Bul 85(2):410–416, doi:10.1037/0033-2909.85.2.410

Knight RL, Walton WE, O’Meara GF, Reisen WK, Wass R (2003) Strategies foreffective mosquito control in constructed treatment wetlands. Ecol Eng21:211–232

Koivisto S, Arnér M, Kautsky N (1997) Does cadmium pollution change trophicinteractions in rockpool food webs? Envir Tox Chem 16:1330–1336

Kumar R, Muhid P, Dahms HU, Tseng LC, Hwang JS (2008) Potential of threeaquatic predators to control mosquitoes in the presence of alternative prey:a comparative experimental assessment. Mar Freshwater Res 59(9):817–35

Kweka EJ, Zhou G, Gilbreath TM, Afrane Y, Nyindo M, Githeko AK, Yan G (2011)Predation efficiency of Anopheles gambiae larvae by aquatic predators inwestern Kenya highlands. Parasites & Vectors 4:128

Kweka EJ, Kamau L, Munga S, Lee M, Githeko AK, Yan G (2013) A first report ofAnopheles funestus sibling species in western Kenya highlands. Acta Trop128:158–61

Lawrence C (2007) The husbandry of zebrafish (Danio rerio): a review.Aquaculture 269:1–20

Matano A (2013) Effects of Land Use on Water Quality and Aquatic Biota of theMara River, PhD thesis. Maseno University, Kenya

Mathooko JM (1998) Mayfly diversity in East Africa. Afr J Ecol 36:368–370

Mati BM, Mutie S, Home P, Mtalo F, Gadain H (2005) Land Use Changes in theTransboundary Mara Basin: A Threat to Pristine Wildlife Sanctuaries in EastAfrica. A Paper presentation at the 8th International River Symposium,Brisbane, Australia

Mati BM, Mutie SM, Gadain H, Home P, Mtalo PF (2008) Impacts of land-use/coverchanges on the hydrology of the transboundary Mara River, Kenya/Tanzania,Lakes and Reservoirs. Res Manag 13:169–177

Matthys B, N’Goran EK, Kone M, Koudou BG, Vounatsou P, Cisse G, Tschannen AB,Tanner M, Utzinger J (2006) Urban agricultural land use and characterizationof mosquito larval habitats in a medium-sized town of Cote d’Ivoire. J VectorEcol 31:319–333

McPeek MA, Miller TE (1996) Evolutionary biology and community ecology.Ecol 77:1319–1320

Medlock JM, Snow KR (2008) Natural predators and parasites of Britishmosquitoes– a review. Eur Mosq Bull. 25:1–11

Merritt RW, Cummins KW (1996) An Introduction to the Aquatic Insects of NorthAmerica, 3rd edn. Kendall Hunt, Dubuque

Minakawa N, Mutero C, Githure JI, Beier JC, Yan G (1999) Spatial distribution andhabitat characterisation of anopheline mosquito larvae in western Kenya.Am J Trop Med Hyg 61(6):1010–1016

Minakawa N, Futami K, Sonye G, Akweywa P, Kaneko S (2007) Predatory capacityof a shorefly, Ochthera chalybescens, on malaria vectors. Malar J6:104, PubMed: 17683604

Minakawa N, Sonye G, Dida GO, Futami K, Kaneko S (2008) Recent reduction inthe water level of Lake Victoria has created more habitats for Anophelesfunestus. Malar J 7:119

Minakawa N, Dida GO, Sonye GO, Futami K, Njenga SM (2012) Malaria vectors inLake Victoria and adjacent habitats in Western Kenya. PLoS One 7:e32725

Mogi M (2007) Insects and other invertebrate predators. J Am Mos Cont Asso 23(suppl 3):93–109

Munga S, Vulule J, Kweka EJ (2013) Response of Anopheles gambiae s.l. (Diptera:Culicidae) to larval habitat age in western Kenya highlands. Parasites &Vectors 6:13

Murdoch WW, Scott MA, Ebsworth P (1984) Effects of general predator,Notonecta (Hemiptera) upon a freshwater community. J Anim Ecol 53:791–808

Mutie SM, Mati B, Home P, Gadain H, Gathenya J (2006) Evaluating Land UseChange Effects on River Flow Using USGS Geospatial Stream Flow Model inMara River Basin, Kenya. Center for Remote Sensing of Land Surfaces, Bonn

Muturi J, Mwangangi J, Shililu J, Jacob G, Mbogo C, Githure J, Novak J (2008)Environmental factors associated with the distribution of Anopheles arabiensisand Culex quinquefasciatus in a rice agro-ecosystem in Mwea, Kenya. J VectorEcol 33(1):56–63

Mwangangi JM, Mbogo CM, Muturi EJ, Nzovu JG, Githure JI, Yan G (2007) Spatialdistribution and habitat characterisation of Anopheles larvae along theKenyan coast. J Vector Borne Dis 44:44–51

Nelson FRS (1979) Comparative predatory potential and asexual reproduction ofsectioned Dugesia dorotocephala as they relate to biological control ofmosquito vectors. Environ Ent 8:679–81

Nilson A (ed) (1996) Aquatic Insects of North Europe; A Taxonomic Handbook,vol 1: Ephemeroptera-Plecoptera-Heterotera-Neuroptera-Megaloptera-Cole-optera-Trichoptera-Lepidoptera. Apollo Books, Stenstrup

Nilson A (ed) (1997) Aquatic insects of north Europe; a taxonomic handbook, vol. 2:Odonata-Diptera. Apollo Books, Stenstrup

Ohba S, Nakasuji F (2006) Dietary items of predacious aquatic bugs (Nepoidea:Heteroptera) in Japanese wetlands. Limn 7:41–43

Ohba S, Kawada H, Dida G, Juma D, Sonye DG, Minakawa N, Takagi M (2010)Predators of Anopheles gambiae sensu lato (Diptera: Culicidae) Larvae inWetlands, Western Kenya: Confirmation by Polymerase Chain ReactionMethod. J Med Entomol 47(5):783–787

Okbah MA, Tayel FTR (1999) Water quality in the coastal area of Alexandria.Bull of the Nat Insti of Oceano and Fisher 25:89–102

Opoku AA, Ansa-Asare OD & Amoako J (2007) The Occurrences and HabitatCharacteristics of Mosquitoes in Accra, Ghana, West Africa. Journal of AppliedEcolology, 11:81–86.

Oyewole IO, Momoh OO, Anyasor GN, Ogunnowo AA, Ibidapo CA, Oduola Oet al (2009) Physico-chemical characteristics of Anopheles breeding sites:Impact on fecundity and progeny Development. Afr J Environ Sci Technol3(12):447–452

Quiroz-Martinez H, Rodriguez-Castro A (2007) Aquatic insects as predators ofmosquito larvae. JAm Mosq Control Assoc 23(2):110–117

Dida et al. SpringerPlus (2015) 4:136 Page 14 of 14

Schlagenhauf-lawlor P, Scott J (2001) Traveller’s Malaria. PMPH Publsher’s, USA,p 155009157

Service MW (1971) Studies on sampling larval populations of the Anophelesgambiae complex. Bull WHO 45:169–180 [PubMed: 5316615]

Service MW (1973) Mortalities of the larvae of the Anopheles gambiae Gilescomplex and detection of predators by the precipitin test. Bull Entomol Res62:359–369

Service MW (1977) Mortalities of the immature stages of species B of the Anophelesgambiae complex in Kenya: comparison between rice fields and temporarypools, identification of predators. J Med Entomol 31 13(4-5):535–545

Shaalan EA, Canyon DV, Reinhold M, Yones WFM, Abdel-Wahab H, Mansour A(2007) A mosquito predator survey in Townsville, Australia and an assessment ofDiplonychus sp. and Anisops sp. predatorial capacity against Culex annulirostrismosquito immatures. J Vector Ecol 32(1):16–21

Shiff C (2002) Integrated approach to malaria control. Clin Microbiol Rev 15:278–93Skourkeas A, Kolyva-Machera F, Maheras P (2010) Estimation of mean maximum

summer and mean minimum winter temperatures over Greece in 2070–2100using statistical downscaling methods. EuroAsian J Sustain Energy Dev Policy2:33–44

Spieles DJ, Mitsch WJ (2000) The effects of season and hydrologic and chemicalloading on nitrate retention in constructed wetlands: A comparison of lowand high nutrient riverine systems. Ecol Eng 14:77–91

Stav G, Blaustein L, Margalith J (1999) Experimental evidence for predation risksensitive oviposition by a mosquito, Culiseta longiareolata. Ecol Entomol24:202–207

Sumba LA, Ogbunugafor CB, Deng AL, Hassanali A (2008) Regulation ofoviposition in Anopheles gambiae s.s.: role of inter-and intra-specific signals,J. Chem. Ecol 34(11):1430–1436

Thomson DJ (1978) Towards a realistic predator–prey model: the effect oftemperature on the functional response and life history of larvae of thedamselfly, Ischnura elegans. J Anim Ecol 47:757–767

Travis J, Keen WH, Juilianna J (1985) The role of relative body size in a predator–preyrelationship between dragonfly naiads and larval anurans. Oikos 45:59–65

Tuno NWO, Minakawa N, Takagi M, Yan G (2005) Survivorship of Anophelesgambiae sensu stricto (Diptera: Culicidae) larvae in Western Kenya highlandforest. J Med Entomol 42(Tuno NW Okeka, N Minakawa, M Takagi, and GYan):270–277

Tuno N, Githeko AK, Nakayama T, Minakawa N, Takagi M, Yan GY (2006) Theassociation between the phytoplankton, Rhopalosolen species (Chlorophyta;Chlorophyceae), and Anopheles gambiae sensu lato (Diptera: Culicidae) larvalabundance in western Kenya. Ecol Res 21:476–482

Tuno N, Githeko A, Yan G, Takagi M (2007) Interspecific variation in diving activityamong Anopheles gambiae Giles, Anopheles arabiensis Patton, and Anophelesfunestus Giles (Diptera: Culicidae) larvae. J Vector Ecol 32:112–117

Verschuren D (1997) Taxonomy and ecology of subfossil Chironomidae (Insecta:Diptera) from Rift Valley lakes in central Kenya. Arch Hydrobiol 107:467–512

Vince S, Valiela I, Backus N (1976) Predation by the salt marsh killifish Fundulusheteroclitus (L.) in relation to prey size and habitat structure: consequencesfor prey distribution and abundance. J Exp Mar Biol Ecol 23:255–266

WHO (2013) World Malaria Report, WHO Fact SheetWilson C, Tisdell C (2001) Why farmers continue to use pesticides despite

environmental, health and sustainability costs. Ecol Econ 39:446–449Witte F, Van Oijen MJP (1990) Taxonomy, ecology and fishery of Lake Victoria

haplochromine trophic groups. Zool verh, Leiden 262:1–47

Submit your manuscript to a journal and benefi t from:

7 Convenient online submission

7 Rigorous peer review

7 Immediate publication on acceptance

7 Open access: articles freely available online

7 High visibility within the fi eld

7 Retaining the copyright to your article

Submit your next manuscript at 7 springeropen.com